TW202223807A - System and method of product recommendation and computer readable medium - Google Patents
System and method of product recommendation and computer readable medium Download PDFInfo
- Publication number
- TW202223807A TW202223807A TW109143669A TW109143669A TW202223807A TW 202223807 A TW202223807 A TW 202223807A TW 109143669 A TW109143669 A TW 109143669A TW 109143669 A TW109143669 A TW 109143669A TW 202223807 A TW202223807 A TW 202223807A
- Authority
- TW
- Taiwan
- Prior art keywords
- product
- user
- record
- records
- recommendation
- Prior art date
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Information Transfer Between Computers (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
Description
本發明係有關一種商品推薦技術,特別係有關一種基於關鍵字或用戶行為之商品推薦系統與方法。 The present invention relates to a commodity recommendation technology, in particular to a commodity recommendation system and method based on keywords or user behavior.
過往企業採取被動式營銷,雖有許多能提供用戶更多便利之服務與產品,但必須等到用戶有需求到達門市時,才能由門市人員為用戶介紹或推薦商品,導致銷售率及企業營收無法大幅度提升,在占領市場時也會落後其他業者,這會對企業的人力、財力、營銷及市場佈署等層面造成不利影響。 In the past, companies adopted passive marketing. Although there are many services and products that can provide users with more convenience, they must wait until the user has a demand to reach the store before the store staff can introduce or recommend products to the user, resulting in the sales rate and corporate revenue cannot be large. If the rate increases, it will also lag behind other players when occupying the market, which will have an adverse impact on the human, financial, marketing and market deployment of the company.
因此,需要一種有效、快速、精確且主動地提供用戶更多便利之服務與產品的技術,以提高商品推薦的效益。 Therefore, there is a need for a technology that can effectively, quickly, accurately and proactively provide users with more convenient services and products, so as to improve the effectiveness of product recommendation.
為解決上述問題,本發明提供一種商品推薦系統,包括:人工智慧互動模組,用於提供對話介面,且通過該對話介面與用戶進行對話互動,以接收該用戶之對話文字;以及關鍵字推薦模組,用於擷取該對話文字中之至 少一關鍵字,計算各該關鍵字之推薦順序,以根據該推薦順序向該用戶提供各該關鍵字之推薦商品。 In order to solve the above problems, the present invention provides a product recommendation system, including: an artificial intelligence interactive module for providing a dialogue interface, and through the dialogue interface to conduct dialogue and interaction with a user to receive the user's dialogue text; and keyword recommendation A module for retrieving up to One keyword is missing, and the recommendation order of each keyword is calculated, so as to provide the user with the recommended products of each keyword according to the recommendation order.
本發明另提供一種商品推薦系統,包括:複數記錄模組,用於蒐集該商品推薦系統之各用戶在網路上的至少一商品相關之購買記錄、廣告點擊記錄及瀏覽記錄;門市資料模組,用於蒐集該商品推薦系統之各該用戶在實體門市中之行為記錄及商品互動記錄;以及混合推薦模組,用於在該商品推薦系統之該等用戶中選定目標用戶,以根據該購買記錄、該廣告點擊記錄、該瀏覽記錄、該行為記錄及該商品互動記錄,向該目標用戶提供推薦商品。 The present invention further provides a product recommendation system, comprising: a plurality of record modules for collecting purchase records, advertisement click records and browsing records related to at least one product on the Internet of each user of the product recommendation system; a store data module, It is used to collect the behavior records and product interaction records of the users of the product recommendation system in the physical store; and the hybrid recommendation module is used to select target users among the users of the product recommendation system, and based on the purchase records , the advertisement click record, the browsing record, the behavior record and the product interaction record, and provide recommended products to the target user.
本發明又提供一種商品推薦方法,包括:提供對話介面;通過該對話介面與用戶進行對話互動,以接收該用戶之對話文字;擷取該對話文字中之至少一關鍵字,以計算各該關鍵字之推薦順序;以及根據該推薦順序向該用戶提供各該關鍵字之推薦商品。 The present invention further provides a product recommendation method, comprising: providing a dialogue interface; interacting with a user through the dialogue interface to receive the user's dialogue text; retrieving at least one keyword in the dialogue text to calculate the key The recommended order of the words; and according to the recommended order, the user is provided with the recommended products of each of the keywords.
本發明再提供一種商品推薦方法,包括:蒐集各用戶在網路上的至少一商品相關之購買記錄、廣告點擊記錄及瀏覽記錄;蒐集各該用戶在實體門市中之行為記錄及商品互動記錄;在該等用戶中選定目標用戶,以根據該購買記錄、該廣告點擊記錄、該瀏覽記錄、該行為記錄及該商品互動記錄,向該目標用戶提供推薦商品。 The present invention further provides a product recommendation method, which includes: collecting purchase records, advertisement click records and browsing records related to at least one product on the Internet; collecting behavior records and product interaction records of the users in physical stores; A target user is selected among the users to provide recommended products to the target user according to the purchase record, the advertisement click record, the browsing record, the behavior record and the product interaction record.
本發明復提供一種電腦可讀媒介,應用於計算裝置或電腦中,係儲存有指令,以執行上述之商品推薦方法。 The present invention further provides a computer-readable medium, which is applied to a computing device or a computer and stores instructions for executing the above-mentioned method for recommending products.
本發明之商品推薦系統與方法及電腦可讀媒介係改善整體銷售方式,使門市人員不再被動式推薦商品給予用戶,可改為主動式推薦。此外,透過清楚明白用戶資訊內容以及用戶推薦內容,可大幅度地提升整體銷售業 績。再者,若用戶無法到實體門市時,則透過系統推播商品推薦資訊給用戶,將原來的被動營銷改為主動營銷,可快速發現目標用戶、快速占領細分市場、並改善銷售策略。 The product recommendation system and method and the computer-readable medium of the present invention improve the overall sales method, so that store staff can no longer passively recommend products to users, but can instead actively recommend products. In addition, by clearly understanding the content of user information and user recommendations, the overall sales business can be greatly improved achievement. Furthermore, if the user cannot go to the physical store, the system pushes product recommendation information to the user, changing the original passive marketing to active marketing, which can quickly find target users, quickly occupy market segments, and improve sales strategies.
100:商品推薦系統 100: Product Recommendation System
101:用戶 101: User
102:行動裝置 102: Mobile Devices
103:電子裝置 103: Electronic Devices
104:實體門市 104: Physical store
110:人工智慧互動模組 110: Artificial Intelligence Interactive Module
121:購買記錄模組 121: Purchase Recording Module
122:廣告點擊模組 122:Ad click module
123:瀏覽記錄模組 123:Browsing record module
124:定位模組 124: Positioning module
125:生物特徵模組 125: Biometric Modules
126:門市資料模組 126: Store data module
130,140:資料庫 130,140:Database
150:混合推薦模組 150: Hybrid recommendation mod
151:關鍵字推薦模組 151:Keyword Recommendation Module
152:用戶相似度推薦模組 152: User similarity recommendation module
153:商品相似度推薦模組 153: Commodity similarity recommendation module
401~404:用戶 401~404: User
451~454:商品 451~454: Commodities
601:用戶 601: User
651~653:商品 651~653: Commodities
S210~S270,S310~S360,S510~S560:方法步驟 S210~S270, S310~S360, S510~S560: method steps
圖1為根據本發明一實施例之一種商品推薦系統的示意方塊圖。 FIG. 1 is a schematic block diagram of a product recommendation system according to an embodiment of the present invention.
圖2為根據本發明一實施例之一種商品推薦方法的關鍵字推薦流程圖。 FIG. 2 is a flowchart of keyword recommendation of a product recommendation method according to an embodiment of the present invention.
圖3為根據本發明一實施例之一種商品推薦方法的用戶相似度推薦流程圖。 FIG. 3 is a flowchart of user similarity recommendation according to a product recommendation method according to an embodiment of the present invention.
圖4為根據本發明一實施例之一種商品推薦方法的用戶相似度推薦範例圖。 FIG. 4 is a diagram illustrating an example of user similarity recommendation according to a product recommendation method according to an embodiment of the present invention.
圖5為根據本發明一實施例之一種商品推薦方法的商品相似度推薦流程圖。 FIG. 5 is a flow chart of recommending product similarity in a product recommending method according to an embodiment of the present invention.
圖6為根據本發明一實施例之一種商品推薦方法的商品相似度推薦範例圖。 FIG. 6 is a diagram illustrating an example of product similarity recommendation according to a product recommendation method according to an embodiment of the present invention.
以下藉由特定的具體實施例說明本發明之實施方式,在本技術領域具有通常知識者可由本說明書所揭示之內容輕易地瞭解本發明之其他優點及功效。 The embodiments of the present invention are described below by means of specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification.
圖1為根據本發明一實施例之一種商品推薦系統100的示意方塊圖。商品推薦系統100包括人工智慧互動模組110、購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123、定位模組124、生物特徵模組125、門市資料模組126、資料庫130及140、以及混合推薦模組150,其中,混合推薦模組150包括關鍵字推薦模組151、用戶相似度推薦模組152、以及商品相似度推薦模組153。資料庫130通訊連接人工智慧互動模組110及關鍵字推薦模組151。資料庫140通訊連接購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123、定位模組124、生物特徵模組125、門市資料模組126、用戶相似度推薦模組152、以及商品相似度推薦模組153。
FIG. 1 is a schematic block diagram of a
圖1中之商品推薦系統100的各模組均可為軟體、硬體或韌體;若為硬體,則可為具有資料處理與運算能力之處理單元、處理器、電腦或伺服器;若為軟體或韌體,則可包括處理單元、處理器、電腦或伺服器可執行之指令。圖1中之商品推薦系統100的各模組及各資料庫可整合於同一硬體裝置中,或分散建置於複數硬體裝置中。
Each module of the
商品推薦系統100可隸屬於兼營網路門市及實體門市之電信公司或商家,用於向用戶推薦商品。上述商品為可購買之實體產品(例如各種電子產品或消費產品)或可申辦之服務(例如電信服務、醫療服務或保險服務)。人工智慧互動模組110、購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123、定位模組124、生物特徵模組125、以及門市資料模組126用於自用戶之行動裝置及電腦等電子裝置(例如用戶101之行動裝置102及電子裝置103)蒐集各種資料,並將該等資料分別存入資料庫130及140,關鍵字推薦模
組151、用戶相似度推薦模組152、以及商品相似度推薦模組153則用於根據該等資料分別執行如圖2、圖3及圖5所示之商品推薦方法。
The
圖2為根據本發明一實施例之一種商品推薦方法的關鍵字推薦流程圖。 FIG. 2 is a flowchart of keyword recommendation of a product recommendation method according to an embodiment of the present invention.
首先,在步驟S210,用戶101可使用行動裝置102或電子裝置103開啟人工智慧互動模組110所提供之對話介面,利用文字輸入,與人工智慧互動模組110進行對話互動,以瞭解電信公司或商家提供之各種商品。人工智慧互動模組110藉由已訓練之人工智慧模型並通過對話介面與用戶101進行對話互動。
First, in step S210, the
在步驟S220,人工智慧互動模組110通過對話介面接收用戶101在對話互動過程中所輸入之文字(以下簡稱為對話文字),以將對話文字儲存至資料庫130。
In step S220 , the artificial
接著,在步驟S230,關鍵字推薦模組151自資料庫130取得用戶101之對話文字,再過濾掉對話文字中之標點符號及表情符號等特殊符號,並過濾掉對話文字中之語助詞及感嘆詞等干擾詞。
Next, in step S230, the
在步驟S240,關鍵字推薦模組151擷取過濾後之對話文字中的至少一關鍵字,再以下列公式計算各關鍵字i之出現頻率tf i 。
In step S240, the
用戶之對話文字可包括複數文句,其中,用戶在對話互動過程中每次點擊對話介面之傳送鍵所發送之文字為一個文句。上述公式中,n i 為用戶101之對話文字中包含關鍵字i之文句數量,而N為用戶101之對話文字中
之全部文句數量。換言之,各關鍵字i之出現頻率tf i 係根據各關鍵字i在用戶101之對話文字中的出現比例而計算產生。
The dialogue text of the user may include plural text sentences, wherein the text sent by the user each time the user clicks the transmission button of the dialogue interface during the dialogue interaction process is one text sentence. In the above formula, n i is the number of sentences containing the keyword i in the dialogue text of the
接著,在步驟S250,關鍵字推薦模組151以下列公式計算過濾後之對話文字中的各關鍵字i之重要性idf i 。
Next, in step S250, the
用戶之對話文字中之文句可劃分為複數節段,且各該節段均包括複數文句,各該節段所包括之文句數量可以相同,也可以不相同。各該節段可用檔案等任何形式儲存。上述公式中,D為資料庫130中儲存之所有用戶(不限於用戶101,而是電信公司或商家之所有用戶,該等用戶亦可視為商品推薦系統100之所有用戶)的所有對話文字之節段總數量,而d i 為資料庫130中儲存之所有用戶的所有對話文字中,包含關鍵字i之節段總數量。換言之,各關鍵字i之重要性idf i 係根據各關鍵字i在所有用戶之對話文字中的出現比例而計算產生。
The sentences in the dialogue text of the user can be divided into plural sections, and each section includes plural sentences, and the number of sentences included in each section can be the same or different. Each of the sections can be stored in any form such as a file. In the above formula, D is the section of all dialogue texts of all users stored in the database 130 (not limited to
接著,在步驟S260,關鍵字推薦模組151以下列公式計算過濾後之對話文字中的各關鍵字i之推薦順序R i 。
Next, in step S260, the
R i =tf i ×idf i R i = tf i × idf i
接著,在步驟S270,關鍵字推薦模組151根據各關鍵字i之推薦順序R i 向用戶101提供並推薦至少一個關鍵字i之相關商品。詳言之,在對話文字的複數關鍵字中,當某一關鍵字i之推薦順序R i 之數值愈高,則該關鍵字i之相關商品之推薦順序愈優先。關鍵字推薦模組151可將上述相關商品之
推薦資訊提供給門市人員參考,亦可將上述相關商品之推薦資訊傳送至用戶101之行動裝置102及/或電子裝置103。
Next, in step S270, the
例如,用戶101在與人工智慧互動模組110之對話互動過程中輸入「我想了解最新的手機」,並且在後續之對話互動過程中多次提到關鍵字「手機」,則經過上述之公式計算後,關鍵字推薦模組151可判斷用戶101可能對手機商品有興趣,而傳送手機商品之推薦資訊至用戶101的行動裝置102及/或電子裝置103。
For example, when the
圖3為根據本發明一實施例之一種商品推薦方法的用戶相似度推薦流程圖。 FIG. 3 is a flowchart of user similarity recommendation according to a product recommendation method according to an embodiment of the present invention.
首先,在步驟S310,購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123、以及門市資料模組126分別蒐集電信公司或商家的每一用戶之購買記錄、廣告點擊記錄、瀏覽記錄及門市資料,並將這些記錄及資料存入資料庫140。詳言之,每一用戶之購買記錄係該用戶在網路上購買或申辦該電信公司或商家的商品之記錄。每一用戶之廣告點擊記錄係該用戶在網站或行動裝置應用程式中點擊該電信公司或商家的商品廣告之記錄。每一用戶之瀏覽記錄包括該用戶在網路上搜尋過及/或瀏覽過的與該電信公司或商家之商品相關的網頁之記錄。
First, in step S310, the
每一用戶之門市資料包括該用戶之顧客屬性、以及該用戶在該電信公司或商家的實體門市(例如實體門市104)中之行為記錄及商品互動記錄。該顧客屬性可包括該用戶之性別、年齡、收入及興趣等基本資料,可由該用戶在該電信公司或商家之網路門市或實體門市填寫。該行為記錄可包括該用戶在該電信公司或商家之實體門市中查詢商品訊息、購買商品及/或申辦商品的 記錄。在實體門市中,該行為記錄可由服務人員或自動服務設備記錄。該商品互動記錄可包括該用戶與實體門市中陳列之商品互動的記錄,例如,該用戶試用過哪些商品,或曾觀看哪些商品且觀看之持續時間超過預設值,上述之商品互動表示該用戶可能對試用或觀看的商品感興趣。 The store data of each user includes the user's customer attributes, as well as the user's behavior records and product interaction records in the telecommunication company's or merchant's physical stores (eg, the physical store 104 ). The customer attributes may include basic information such as the user's gender, age, income, and interests, which may be filled in by the user at the online or physical stores of the telecommunications company or business. The behavior record may include the user's inquiry of commodity information, purchase of commodities and/or application for commodities in the physical stores of the telecommunication company or merchant. Record. In brick-and-mortar stores, the behavior record can be recorded by service personnel or automated service equipment. The product interaction record may include a record of the user's interaction with the products displayed in the physical store, for example, which products the user has tried, or which products the user has watched and the viewing duration exceeds a preset value. The above-mentioned product interaction means that the user Items that may be of interest to try or watch.
在一實施例中,實體門市中係裝設攝影機及錄音機,用於收錄用戶在實體門市中之影像與聲音。門市資料模組126可自實體門市之攝影機及錄音機取得用戶在實體門市中之影像與聲音,並將該等影像與聲音存入資料庫140。門市資料模組126可分析該等影像與聲音以擷取用戶在實體門市中之商品互動記錄,再將該商品互動記錄存入資料庫140。此外,門市資料模組126亦可分析該等影像與聲音以擷取用戶在實體門市中之行為記錄,再將該行為記錄存入資料庫140,以做為服務人員或自動服務設備之記錄之補充。
In one embodiment, a camera and a sound recorder are installed in the physical store for recording images and sounds of the user in the physical store. The
為了在該等影像與聲音中辨識各用戶,實體門市之服務人員或自動服務設備可記錄各用戶到達實體門市之時間與地點(例如,哪一家實體門市),生物特徵模組125即可根據該時間與地點取得各用戶在該等影像與聲音中出現之片段,且自該等片段擷取各用戶之聲紋、筆跡、臉型與步態等生物特徵,再將各用戶之生物特徵存入資料庫140。門市資料模組126亦可根據該時間與地點取得各用戶在該等影像與聲音中出現之片段,再根據該等生物特徵在該等片段中辨識各用戶,並分析該等片段,以擷取各用戶之商品互動記錄及行為記錄。
In order to identify each user in these images and sounds, the service personnel or automatic service equipment of the physical store can record the time and place (for example, which physical store) each user arrives at the physical store, and the
另外,定位模組124用於取得用戶之位置資訊。若商品推薦系統100隸屬於電信公司,則定位模組124可根據該電信公司之基地台所接收到的用戶之行動裝置的信號強度進行三角定位運算,以取得用戶之位置資訊,並
將此位置資訊存入資料庫140。或者,無論商品推薦系統100隸屬於電信公司或商家,定位模組124亦可通過用戶授權自用戶之行動裝置取得用戶之位置資訊,並將此位置資訊存入資料庫140。該位置資訊可做為實體門市之服務人員或自動服務設備所記錄的各用戶到達實體門市之時間與地點的補充,換言之,生物特徵模組125及門市資料模組126可根據該位置資訊取得各用戶在該等影像與聲音中出現之片段,進而擷取各用戶之生物特徵、商品互動記錄及行為記錄。
In addition, the
接著,在步驟S320,用戶相似度推薦模組152在電信公司或商家之所有用戶中選定目標用戶u,並選定待推薦商品i。例如,目標用戶u可為最近剛購買或申辦商品之用戶,用戶相似度推薦模組152可自購買記錄模組121提供之購買記錄得知有哪些用戶最近剛購買或申辦商品。或者,目標用戶u可為促銷活動所針對之用戶。待推薦商品i可為促銷活動所針對之商品。
Next, in step S320, the user
接著,在步驟S330至S360,用戶相似度推薦模組152自資料庫140取得購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123及門市資料模組126所提供之購買記錄、廣告點擊記錄、瀏覽記錄及門市資料,將這些記錄與資料輸入協同過濾(Collaborative Filtering)模型過濾,並透過用戶相似度矩陣(User-User Similarity Matrix)找出與目標用戶u相似的用戶集合後,將相似用戶最常購買的商品推薦給目標用戶u,其演算法詳述如下。
Next, in steps S330 to S360, the user
在步驟S330,用戶相似度推薦模組152產生與目標用戶u興趣相似之用戶的集合S(u)。在一實施例中,用戶相似度推薦模組152可用下列之雅卡爾相似度公式(Jaccard Index)判斷兩用戶α及β之興趣是否相似。
In step S330, the user
上述之雅卡爾相似度公式中,A及B分別為用戶α及β有過興趣行為的商品之集合,其中,興趣行為係指用戶購買或申辦過某商品、點擊過該商品之廣告、搜尋或瀏覽過該商品相關之網頁、或在實體門市中曾有該商品相關之行為記錄或商品互動記錄。用戶相似度推薦模組152可根據購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123及門市資料模組126所提供之購買記錄、廣告點擊記錄、瀏覽記錄及門市資料得知任一用戶是否對任一商品有過上述之興趣行為,進而產生任一用戶有過興趣行為的商品之集合。若雅卡爾相似度公式計算所得之興趣相似度J(α,β)大於預設值(例如0.5),則判定用戶α及β之興趣相似。用戶相似度推薦模組152可將目標用戶u及每一其他用戶有過興趣行為的商品之集合代入雅卡爾相似度公式,以計算兩用戶之興趣相似度,並將使該興趣相似度大於該預設值之其他用戶加入集合S(u)。
In the above Jacquard similarity formula, A and B are the collections of commodities that users α and β have had interest behaviors, respectively, where the interest behaviors refer to the fact that the user has purchased or applied for a certain commodity, clicked on the advertisement, searched or searched for the commodity. Browsing the webpage related to the product, or having a record of the product-related behavior or product interaction in the physical store. The user
在步驟S340,用戶相似度推薦模組152產生對待推薦商品i有過上述興趣行為之用戶的集合N(i)。
In step S340, the user
在步驟S350,用戶相似度推薦模組152以下列之用戶相似度矩陣公式計算待推薦商品i對於目標用戶u之推薦分數ρ(u,i)。
In step S350, the user
上述公式中,v為集合S(u)與N(i)之交集中之每一用戶,w uv 為目標用戶u和用戶v之興趣相似度,該興趣相似度可用上述之雅卡爾相似度公式計算,r vi 則為用戶v對於待推薦商品i之興趣度。興趣度r vi 和用戶v對於待推薦商品i之興趣行為相關,詳言之,可為上述之每一種興趣行為設定一個 對應的權重值,再將用戶v對於待推薦商品i曾有過之每一種興趣行為所對應的權重值相加,即可得到興趣度r vi 。 In the above formula, v is each user in the intersection of the set S ( u ) and N ( i ), w uv is the interest similarity between the target user u and user v , and the interest similarity can use the above-mentioned Jacquard similarity formula Calculated, r vi is the degree of interest of user v to the item i to be recommended. The degree of interest r vi is related to the interest behavior of user v for the product i to be recommended . The interest degree r vi can be obtained by adding the weight values corresponding to an interest behavior.
在步驟S360,用戶相似度推薦模組152根據推薦分數ρ(u,i)判斷是否對目標用戶u推薦並提供待推薦商品i。詳言之,若推薦分數ρ(u,i)大於預設值(例如0.25),則用戶相似度推薦模組152對目標用戶u推薦待推薦商品i。用戶相似度推薦模組152可將待推薦商品i之推薦資訊提供給門市人員參考,亦可將待推薦商品i之推薦資訊傳送至目標用戶u之行動裝置及或電子裝置。反之,若推薦分數ρ(u,i)小於或等於該預設值,則用戶相似度推薦模組152不對目標用戶u推薦待推薦商品i,用戶相似度推薦模組152可選定另一待推薦商品再重複上述步驟。
In step S360, the user
圖4為根據本發明一實施例之一種商品推薦方法的用戶相似度推薦範例圖,其中之實線表示「購買」,虛線則表示「推薦」。如圖4所示,用戶401購買了商品451、452及454,用戶402購買了商品454,用戶403購買了商品453,且目標用戶404購買了商品451及452。經過上述流程,用戶相似度推薦模組152判定用戶401與目標用戶404之興趣相似,且選定用戶401購買過且目標用戶404未購買之商品454為待推薦商品,然後,因待推薦商品454對於目標用戶404之推薦分數大於預設值,故用戶相似度推薦模組152對目標用戶404推薦商品454。
FIG. 4 is an exemplary diagram of user similarity recommendation according to a product recommendation method according to an embodiment of the present invention, wherein the solid line represents "purchase", and the dotted line represents "recommendation". As shown in FIG. 4 ,
圖5為根據本發明一實施例之一種商品推薦方法的商品相似度推薦流程圖。 FIG. 5 is a flow chart of recommending product similarity in a product recommending method according to an embodiment of the present invention.
首先,在步驟S510,購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123、以及門市資料模組126分別蒐集電信公司或商家的每一用戶
之購買記錄、廣告點擊記錄、瀏覽記錄及門市資料,並將這些記錄及資料存入資料庫140。步驟S510和圖3中之步驟S310相同,細節不再贅述。
First, in step S510, the
接著,在步驟S520,商品相似度推薦模組153在電信公司或商家之所有用戶中選定目標用戶u。
Next, in step S520, the commodity
接著,在步驟S530至S560,商品相似度推薦模組153自資料庫140取得購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123及門市資料模組126所提供之購買記錄、廣告點擊記錄、瀏覽記錄及門市資料,將這些記錄與資料輸入協同過濾(Collaborative Filtering)模型過濾,並透過商品相似度矩陣(Item-Item Similarity Matrix)找出與目標用戶u近期瀏覽或購買之商品相似的商品之集合後,將相似商品推薦給目標用戶u,其演算法詳述如下。
Next, in steps S530 to S560, the product
在步驟S530,商品相似度推薦模組153產生目標用戶u有過興趣行為之商品的集合M(u)。同上所述,商品相似度推薦模組153可根據購買記錄模組121、廣告點擊模組122、瀏覽記錄模組123及門市資料模組126所提供之購買記錄、廣告點擊記錄、瀏覽記錄及門市資料得知任一用戶是否對任一商品有過興趣行為,進而產生任一用戶有過興趣行為之商品的集合。
In step S530, the commodity
在步驟S540,商品相似度推薦模組153產生和目標用戶u已購買之商品j相似之商品的集合S(j)。在此,兩商品相似係指兩商品具有共同之標籤。詳言之,每一商品均可具有複數標籤,用以標示該商品之品牌、產地、是否為有機產品等特徵,若兩商品之全部標籤中,相同標籤之數量除以全部標籤之數量所得的數值大於預設值(例如0.5),則商品相似度推薦模組153判定該兩商品為相似,據此方式,商品相似度推薦模組153可產生和目標用戶u已購買之商品j相似之商品的集合S(j)。
In step S540, the product
在另一實施例中,上述商品j亦可為目標用戶u曾有過興趣行為(例如瀏覽或購買等)之商品,故集合S(j)亦可為和目標用戶u曾有過興趣行為之商品j相似之商品的集合。 In another embodiment, the above-mentioned commodity j can also be a commodity that the target user u has had an interest behavior (such as browsing or purchasing, etc.), so the set S ( j ) can also be a commodity that the target user u has had an interest behavior in the past. A collection of commodities that are similar to commodity j .
在步驟S550,商品相似度推薦模組153以下列之商品相似度矩陣公式計算待推薦商品之推薦分數P(u,i)。
In step S550, the product
上述公式中,商品i為集合M(u)及集合S(j)之交集中之每一商品,w ij 為商品i及商品j之間的相似度,r ui 則為目標用戶u對於商品i之興趣度。商品相似度推薦模組153可用下列公式計算商品i及商品j之間的相似度w ij 。興趣度r ui 之計算可比照前述之興趣度r vi 。
In the above formula, commodity i is each commodity in the intersection of set M ( u ) and set S ( j ), w ij is the similarity between commodity i and commodity j , r ui is the target user u for commodity i of interest. The commodity
上述公式中,N(i)及N(j)分別為對商品i及商品j有過興趣行為之用戶的集合。 In the above formula, N ( i ) and N ( j ) are the sets of users who have been interested in commodity i and commodity j , respectively.
在步驟S560,商品相似度推薦模組153根據推薦分數P(u,i)判斷是否對目標用戶u提供推薦商品i。詳言之,若推薦分數P(u,i)大於預設值(例如0.25),則商品相似度推薦模組153對目標用戶u推薦集合M(u)及集合S(j)之交集中之每一商品i。商品相似度推薦模組153可將商品i之推薦資訊提供給門市人員參考,亦可將商品i之推薦資訊傳送至目標用戶u之行動裝置及/或電子裝置。反之,若推薦分數P(u,i)小於或等於該預設值,則商品相似度推薦模組153不對目標用戶u推薦商品i。
In step S560, the product
圖6為根據本發明一實施例之一種商品推薦方法的商品相似度推薦範例圖,其中之實線表示「購買」,虛線則表示「推薦」。如圖6所示,目標用戶601購買了商品651及652。商品相似度推薦模組153經過上述運算後,向目標用戶601推薦和商品651及652相似之商品653。例如,商品651及652分別為電腦和鍵盤,商品653為滑鼠。
FIG. 6 is a diagram illustrating an example of product similarity recommendation according to a product recommendation method according to an embodiment of the present invention, wherein the solid line represents “purchase”, and the dotted line represents “recommendation”. As shown in FIG. 6 , the
此外,本發明還揭示一種電腦可讀媒介,係應用於具有處理器(例如,CPU、GPU等)及/或記憶體的計算裝置或電腦中,且儲存有指令,並可利用此計算裝置或電腦透過處理器及/或記憶體執行此電腦可讀媒介,以於執行此電腦可讀媒介時執行上述之方法及各步驟。 In addition, the present invention also discloses a computer-readable medium, which is applied to a computing device or computer having a processor (eg, CPU, GPU, etc.) and/or memory, and stores instructions, and can utilize the computing device or computer. The computer executes the computer-readable medium through a processor and/or a memory, so as to execute the above-mentioned methods and steps when executing the computer-readable medium.
綜上所述,本發明提出之商品推薦系統與方法及電腦可讀媒介具有下列特點及功效: To sum up, the product recommendation system and method and the computer-readable medium proposed by the present invention have the following features and effects:
1.人工智慧互動模組可分析用戶語意之意圖,透過對話關鍵字分析出用戶可能感興趣之商品及可能所需之服務資訊,本發明之商品推薦系統與方法可據此推薦商品,藉由此方式提高銷售商品的成功率。 1. The artificial intelligence interactive module can analyze the semantic intention of the user, and analyze the products that the user may be interested in and the service information that the user may need through the dialogue keywords. The product recommendation system and method of the present invention can recommend products accordingly. This method increases the success rate of selling products.
2.倘若客戶曾購買或申辦商品,本發明之商品推薦系統與方法可使用用戶之購買記錄、廣告點擊記錄、瀏覽記錄、以及門市資料分析計算用戶間與商品間之相似度,推薦相似用戶所購買的商品或推薦與目前商品最相似的其他商品,而進行交叉銷售(cross-selling),再轉換成興趣相似的用戶群組後,查看他們最常購買的商品,推薦給目前鎖定的目標用戶。 2. If the customer has purchased or applied for a product, the product recommendation system and method of the present invention can use the user's purchase record, advertisement click record, browsing record, and store data to analyze and calculate the similarity between users and products, and recommend similar users. The purchased products or other products that are most similar to the current products are recommended for cross-selling, and after being converted into a user group with similar interests, the products they buy most often are viewed and recommended to the currently locked target users. .
3.本發明之商品推薦系統與方法可取得用戶之位置資訊,藉以分析用戶過往的地點與用戶可能消費店家,結合用戶之瀏覽記錄及生物特徵等數據,找出與現在瀏覽的商品最相似的商品群組,再推薦給用戶。 3. The product recommendation system and method of the present invention can obtain the user's location information, so as to analyze the user's past location and the store where the user may consume, and combine the user's browsing records and biometric data to find the most similar product to the currently browsed product. Product groups, and then recommend them to users.
4.門市人員可自本發明之商品推薦系統與方法得到即時的用戶需求,以避免推薦不適合商品給用戶,而導致用戶體驗欠佳且影響服務品質,反之,可藉由此方式提高商品銷售率,並增加整體營收。 4. Store personnel can obtain real-time user needs from the product recommendation system and method of the present invention, so as to avoid recommending unsuitable products to users, resulting in poor user experience and affecting service quality. On the contrary, the sales rate of products can be improved in this way. , and increase overall revenue.
上述實施形態僅例示性說明本發明之原理及其功效,而非用於限制本發明。任何在本技術領域具有通常知識者均可在不違背本發明之精神及範疇下,對上述實施形態進行修飾與改變。因此,本發明之權利保護範圍,應如後述之申請專利範圍所列。 The above-mentioned embodiments are only used to illustrate the principle and effect of the present invention, but are not intended to limit the present invention. Anyone with ordinary knowledge in the technical field can modify and change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention should be as listed in the patent application scope described later.
100:商品推薦系統 100: Product Recommendation System
101:用戶 101: User
102:行動裝置 102: Mobile Devices
103:電子裝置 103: Electronic Devices
104:實體門市 104: Physical store
110:人工智慧互動模組 110: Artificial Intelligence Interactive Module
121:購買記錄模組 121: Purchase Recording Module
122:廣告點擊模組 122:Ad click module
123:瀏覽記錄模組 123:Browsing record module
124:定位模組 124: Positioning module
125:生物特徵模組 125: Biometric Modules
126:門市資料模組 126: Store data module
130,140:資料庫 130,140:Database
150:混合推薦模組 150: Hybrid recommendation mod
151:關鍵字推薦模組 151:Keyword Recommendation Module
152:用戶相似度推薦模組 152: User similarity recommendation module
153:商品相似度推薦模組 153: Commodity similarity recommendation module
Claims (11)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW109143669A TWI818213B (en) | 2020-12-10 | 2020-12-10 | System and method of product recommendation and computer readable medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW109143669A TWI818213B (en) | 2020-12-10 | 2020-12-10 | System and method of product recommendation and computer readable medium |
Publications (2)
Publication Number | Publication Date |
---|---|
TW202223807A true TW202223807A (en) | 2022-06-16 |
TWI818213B TWI818213B (en) | 2023-10-11 |
Family
ID=83062654
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW109143669A TWI818213B (en) | 2020-12-10 | 2020-12-10 | System and method of product recommendation and computer readable medium |
Country Status (1)
Country | Link |
---|---|
TW (1) | TWI818213B (en) |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107918622B (en) * | 2016-10-10 | 2022-06-03 | 阿里巴巴集团控股有限公司 | Content recommendation and display method, client, server and system |
CN108805622B (en) * | 2018-06-11 | 2021-11-09 | 深圳乐信软件技术有限公司 | Commodity recommendation method, commodity recommendation device, commodity recommendation equipment and storage medium |
CN109345316A (en) * | 2018-10-23 | 2019-02-15 | 出门问问信息科技有限公司 | Information processing method, device, electronic equipment and computer readable storage medium |
CN111178920B (en) * | 2018-11-09 | 2023-07-07 | 阿里巴巴(深圳)技术有限公司 | Commodity object information recommendation method, device and system |
-
2020
- 2020-12-10 TW TW109143669A patent/TWI818213B/en active
Also Published As
Publication number | Publication date |
---|---|
TWI818213B (en) | 2023-10-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10146776B1 (en) | Method and system for mining image searches to associate images with concepts | |
US10129274B2 (en) | Identifying significant anomalous segments of a metrics dataset | |
JP5456412B2 (en) | Community-based advertising word ambiguity removal system and method | |
Shan et al. | Predicting ad click-through rates via feature-based fully coupled interaction tensor factorization | |
US20220036391A1 (en) | Auto-segmentation | |
US20170249389A1 (en) | Sentiment rating system and method | |
WO2017190610A1 (en) | Target user orientation method and device, and computer storage medium | |
US20150379571A1 (en) | Systems and methods for search retargeting using directed distributed query word representations | |
US20150006286A1 (en) | Targeting users based on categorical content interactions | |
US20130226820A1 (en) | Determining advocacy metrics based on user generated content | |
US20150006294A1 (en) | Targeting rules based on previous recommendations | |
US20120066073A1 (en) | User interest analysis systems and methods | |
TW201520936A (en) | User engagement-based contextually-dependent automated pricing for non-guaranteed delivery | |
US9213733B2 (en) | Computerized internet search system and method | |
US20070255621A1 (en) | Advertisement generation and optimization | |
CN105431875A (en) | Search engine marketing optimizer | |
WO2011019524A2 (en) | Systems and methods for providing targeted content to a network user | |
KR101935937B1 (en) | System for providing advertisement information using opinion mining and a method thereof | |
TWM605342U (en) | Intelligent recommendation system for financial product | |
Chen et al. | Social opinion mining for supporting buyers’ complex decision making: exploratory user study and algorithm comparison | |
US20140074608A1 (en) | Rapid identification of search terms that surge in response to current events | |
CN111429214B (en) | Transaction data-based buyer and seller matching method and device | |
US20150310487A1 (en) | Systems and methods for commercial query suggestion | |
Yamaguchi et al. | Web advertising recommender system based on estimating users' latent interests | |
TWI818213B (en) | System and method of product recommendation and computer readable medium |